Oral Nimodipine in Aneurysmal Subarachnoid Hemorrhage: A Critical Propensity Score-Matched Reevaluation in a Contemporary Cohort

Figures

• 

  Fig. 1. Study flow diagram. Flowchart illustrating patient selection, treatment allocation, and formation of the propensity score–matched cohort. aSAH Aneurysmal subarachnoid hemorrhage, GOS Glasgow Outcome Scale, Nimo nimodipine, Nimo+/Nimo− patients receiving/not receiving oral nimodipine

• 

  Fig. 2. Main analysis in the propensity score-matched cohort. Forest plot showing paired odds ratios with 95% confidence intervals for angiographic vasospasm, delayed cerebral infarction, all infarctions, favorable Glasgow Outcome Scale (GOS) at discharge and at 3 months, and mortality at discharge and at 3 months, comparing treatment with prophylactic oral nimodipine (Nimo+) versus no nimodipine treatment (Nimo−) in the nearest-neighbor propensity score-matched cohort (102 matched pairs). Effect estimates were obtained using conditional logistic regression stratified by matched sets

• 

  Fig. 3. Duration subgroup analyses. Forest plots of paired odds ratios (95% confidence interval) for the same outcomes as Fig. 2 from separate duration-stratified propensity score–matched analyses comparing patients without nimodipine (Nimo−) with those receiving oral nimodipine for < 14 days (Nimo < 14) and ≥ 14 days (Nimo > 14), respectively. Matching was performed using nearest-neighbor propensity score matching (1:2 treated-to-control, without replacement), and treatment effects were estimated using conditional logistic regression stratified by matched sets.

• 

  Fig. 4. Effect modification by clinical severity. Model-based predicted probability of delayed cerebral infarction across ordinal Hunt and Hess grades for patients treated with oral nimodipine versus those not receiving nimodipine, derived from a multivariable logistic regression model including a nimodipine × Hunt–Hess interaction term and adjusting for prespecified covariates. Shaded bands indicate 95% confidence intervals for predicted probabilities; other covariates were held constant at typical reference values during prediction